# eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks architecture

> Architecture diagram of the GitHub repository eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-03). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks
- Repository: https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks

## Overview

This repository supports two research workflows: generating fictional, coordinated multi-channel attack scenarios with Gemini, and evaluating message-level classifiers over a folder of text files. Generation produces scenario documents; the separate training script cleans inputs, invokes Claude and/or GPT-4o, aggregates per-message results, and writes CSV summaries. The README documents the generator workflow; the sampled classifier source shows its cleaning, model calls, and CSV/reporting stages. The checked-in scenario examples are data, but no sampled code establishes that either workflow reads them.

Main entry points: [`model-training.py`](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py) (Classification runner), [`generate_scenarios.py`](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/generate_scenarios.py) (Scenario generator)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_generation["Scenario generation"]
  node_scenario_prompt["Scenario specification"]
  node_scenario_generator["Scenario generator"]
  node_scenario_files["Scenario documents"]
end

subgraph group_classification["Classification pipeline"]
  node_message_files["Message inputs"]
  node_cleaning["Input cleaning<br/>[model-training.py]"]
  node_pipeline["Classification runner<br/>[model-training.py]"]
  node_claude["Claude classifier<br/>[model-training.py]"]
  node_gpt4o["GPT-4o classifier<br/>[model-training.py]"]
end

subgraph group_results["Evaluation results"]
  node_aggregation["Scenario aggregation<br/>[model-training.py]"]
  node_csv_builder["CSV formatting<br/>[model-training.py]"]
  node_csv_output["Results CSV<br/>[model-training.py]"]
  node_summary["Evaluation summary<br/>[model-training.py]"]
end

node_researcher(("Researcher"))
node_gemini["Gemini API"]
node_anthropic_service["Anthropic API"]
node_openai_service["OpenAI API"]

node_researcher -->|"requests scenarios"| node_scenario_generator
node_scenario_generator -->|"uses prompt"| node_scenario_prompt
node_scenario_generator -->|"generates content"| node_gemini
node_gemini -->|"returns content"| node_scenario_generator
node_scenario_generator -->|"writes documents"| node_scenario_files
node_researcher -->|"runs analysis"| node_pipeline
node_pipeline -->|"cleans folder"| node_cleaning
node_pipeline -->|"loads text files"| node_message_files
node_pipeline -->|"classifies messages"| node_claude
node_claude -->|"requests verdict"| node_anthropic_service
node_pipeline -->|"classifies messages"| node_gpt4o
node_gpt4o -->|"requests verdict"| node_openai_service
node_pipeline -->|"collects results"| node_aggregation
node_aggregation -->|"groups results"| node_csv_builder
node_csv_builder -->|"formats rows"| node_csv_output
node_pipeline -->|"saves CSV"| node_csv_output
node_pipeline -->|"prints report"| node_summary

click node_scenario_prompt "https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/generate_scenarios.py"
click node_scenario_generator "https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/generate_scenarios.py"
click node_scenario_files "https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/generate_scenarios.py"
click node_cleaning "https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py"
click node_pipeline "https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py"
click node_claude "https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py"
click node_gpt4o "https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py"
click node_aggregation "https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py"
click node_csv_builder "https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py"
click node_csv_output "https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py"
click node_summary "https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_scenario_prompt,node_scenario_generator,node_scenario_files toneBlue
class node_message_files,node_cleaning,node_pipeline,node_claude,node_gpt4o toneAmber
class node_aggregation,node_csv_builder,node_csv_output,node_summary,node_gemini,node_anthropic_service,node_openai_service toneMint
class node_researcher toneIndigo
```

## Components

### Scenario generation

- [Scenario specification](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/generate_scenarios.py): `generate_scenarios.py`
- [Scenario generator](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/generate_scenarios.py): `generate_scenarios.py`
- [Scenario documents](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/generate_scenarios.py): `generate_scenarios.py`

### Classification pipeline

- **Message inputs**
- [Input cleaning](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py): `model-training.py`
- [Classification runner](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py): `model-training.py`
- [Claude classifier](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py): `model-training.py`
- [GPT-4o classifier](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py): `model-training.py`

### Evaluation results

- [Scenario aggregation](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py): `model-training.py`
- [CSV formatting](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py): `model-training.py`
- [Results CSV](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py): `model-training.py`
- [Evaluation summary](https://github.com/eulando/defending-against-cross-channel-social-engineering-through-agentic-frameworks/blob/main/model-training.py): `model-training.py`

### Other components

- **Researcher**
- **Gemini API**
- **Anthropic API**
- **OpenAI API**

## Connections

- Researcher → Scenario generator: requests scenarios (evidence: `README.md`)
- Scenario generator → Scenario specification: uses prompt (evidence: `generate_scenarios.py`)
- Scenario generator → Gemini API: generates content (evidence: `generate_scenarios.py`)
- Gemini API → Scenario generator: returns content (evidence: `generate_scenarios.py`)
- Scenario generator → Scenario documents: writes documents (evidence: `generate_scenarios.py`)
- Researcher → Classification runner: runs analysis (evidence: `model-training.py`)
- Classification runner → Input cleaning: cleans folder (evidence: `model-training.py`)
- Classification runner → Message inputs: loads text files (evidence: `model-training.py`)
- Classification runner → Claude classifier: classifies messages (evidence: `model-training.py`)
- Claude classifier → Anthropic API: requests verdict (evidence: `model-training.py`)
- Classification runner → GPT-4o classifier: classifies messages (evidence: `model-training.py`)
- GPT-4o classifier → OpenAI API: requests verdict (evidence: `model-training.py`)
- Classification runner → Scenario aggregation: collects results (evidence: `model-training.py`)
- Scenario aggregation → CSV formatting: groups results (evidence: `model-training.py`)
- CSV formatting → Results CSV: formats rows (evidence: `model-training.py`)
- Classification runner → Results CSV: saves CSV (evidence: `model-training.py`)
- Classification runner → Evaluation summary: prints report (evidence: `model-training.py`)
